INDEX

Explanations

names or references related to specific individuals

oai_token-act-pair · gpt-3.5-turbo

names of individuals, particularly in a historical or political context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 rapt

-0.73

 masse

-0.71

 Rapt

-0.70

XY

-0.69

 flyer

-0.69

BDS

-0.68

 pole

-0.68

 pant

-0.67

 prem

-0.63

 Pole

-0.62

POSITIVE LOGITS

ÃŃn

4.22

imeo

1.53

inka

1.49

Ã¡s

1.31

imir

1.28

Ã©n

1.16

hiba

1.12

aldi

1.01

akov

0.99

 winding

0.98

Activations Density 0.044%

names or references related to specific individuals

names of individuals, particularly in a historical or political context

No Comments

No Known Activations